Prophylaxis of Endocarditis 5

(1)

Case Study

An otherwise well 53-year-old man had mitral valve prolapse diagnosed 20 years prior, and had been clinically stable. He presented with an eight- week history of night sweats and a 5-kg weight loss. Approximately one month prior to the onset of symptoms, the patient underwent a dental cleaning and took amoxicillin 2 g, 1 h prior to the procedure. The physical examination revealed a man who appeared well and whose blood pres- sure in the right arm in the sitting position was 118/64 mm Hg with a heart rate of 84 beats per minute (regular). His chest was clear to auscultation and his heart sounds were normal with the exception of a grade 3/6 systolic murmur radiat- ing to the axilla. The peripheral pulses were all palpable and peripheral edema was absent. A blood culture yielded Streptococcus mutans.

A transthoracic echocardiogram revealed significant myxomatous mitral valve disease;

marked thickening of the posterior leaflet with a shaggy appearance and flail segment involving predominantly the middle scallop were seen.

Severe eccentric mitral regurgitation was present.

The left atrium was significantly enlarged. This study was followed up with a transesophageal echocardiogram, which demonstrated that the posterior mitral valve leaflet was diffusely thickened and very redundant. There was severe prolapse of this leaflet. There was at least one small mobile mass at the leaflet tip, but the entire posterior leaflet was thickened and somewhat shaggy.

The findings were consistent with endocarditis.

The S. mutans had a minimal inhibitory con- centration (MIC) to penicillin of 0.008 g/mL.

Since the patient was stable, it was elected to ini- tiate home parenteral antimicrobial therapy with penicillin G, 18 million units per day administered by continuous infusion pump for 4 weeks. The patient had an uneventful course of therapy and underwent an elective mitral valve replacement one year later.

Introduction

Infective endocarditis (IE) is a potentially fatal disease. Even with appropriate antimicrobial treatment, mortality rates range from 10% to 25% [1]; therefore, prevention of disease is very important. Guidelines have been created to estimate which patients with certain risk factors would most benefit from IE prophylaxis.

However, there have been no controlled, clinical trials to demonstrate the protective efficacy of antibiotic regimens in the prophylaxis of IE in humans. Such trials will not likely ever be done for two major reasons: From a study-design perspective, the relative rarity of IE developing after a single transient bacteremic episode would require ≥ 6,000 patients, all with predisposing cardiac disease [2]. Secondly, such a study would also be considered unethical. Therefore, the guidelines that have been devised have been based on the efficacy of IE prophylaxis in animal models, previous antimicrobial susceptibility testing data of the most likely pathogens, pharmacokinetic studies, and studies on the incidence and prophylaxis of procedure-related bacteremias. Thus, the evidence for these recommendations is at the level of expert opinion, the efficacy is not 100%, and the

5

Prophylaxis of Endocarditis

Donald C. Vinh and John M. Embil

47

(2)

changing microbiology of IE may necessitate updated new recommendations.

Pathogenesis and Rationale for Prophylaxis

The fundamental step in the pathogenesis of IE is the development of bacteremia, with subsequent seeding of a previously damaged endocardial surface. Experimental studies suggest that valvular endothelial damage leads to platelet and fibrin deposition and the formation of a nonbacterial thrombotic vegetation. Circulating bacteria can then adhere to these lesions and multiply within the platelet-fibrin complex, leading to an infected vegetation. Dental treatment has traditionally been considered the major cause of the bacteremia that leads to IE [3], mainly because of historical studies that demonstrated a high frequency of bacteremia after various oral invasive procedures, as well as because of previous studies documenting the viridans group streptococci (VGS, the predominant members of the oral microflora) as the leading cause of IE. The initial recognition of a relationship between viridans streptococcal IE and dental procedures is attributed to Horder in 1909 [2,3]. In 1923, Lewis and Grant proposed the hypothesis that abnormally structured heart valves may contribute to the development of IE in healthy adults by trapping and retaining organisms from the transient bacteremia [4]. In 1935, Okell and Elliott, in a series of 138 patients, demonstrated the presence of bacteremia related to tooth extraction; in 64% of the cases, the isolate was a Streptococcus spp. [5]. Another study, published in 1937 by Burket and Burn [6], confirmed the biological plausibility of the oral cavity as the source of bacteremia when they painted the gingival crevices of 90 patients with S. marcescens (which was felt to be non-patho- genic at the time) before dental extraction.

Subsequent to the procedure, the organism was recovered in 20% of the blood cultures. One study demonstrated a “dose-dependent”-like effect, with a significant correlation found between the number of teeth extracted and subsequent positive blood cultures [7]. Thus, it has become well established that bacteremia may occur after dental procedures that compromise mucosal surfaces, especially dental extractions and gingival surgery [8]. This bacteremia, however, is transient, lasting typically no more than 15–30

minutes [9,10], as well as low grade (usually < 100 colony-forming units/mL of blood) [9]. Transient asymptomatic bacteremia also occurs after a variety of other procedures and manipulations, particularly those associated with trauma to the mucous membranes of the respiratory, esophageal, gastrointestinal, and genito-urinary tracts. If the bacteremia following these procedures is a major cause of IE, in theory, maneuvers that decrease the magnitude and/or the duration of this bacteremia could prevent the development of IE in patients at risk for the disease.

Prophylaxis of Experimental Endocarditis

The evidence supporting the use of prophylactic antibiotic regimens in humans derives from its proven efficacy in animal models. Experimental IE has been typically produced in rabbits (e.g., New Zealand white rabbit [11]) or rats (e.g., female Wistar rats [12]) via catheter-induced damage to cardiac valves and subsequent intravenous challenge with various amounts of bacterial inocula. These experimental conditions allowed IE to be more effectively and reliably induced than in other models, with a predictable time of onset, thus facilitating analyses.

Antibiotics are administered at the same or similar weight-based dose as in humans. The experimental IE is followed with serial blood cultures, with eventual sacrifice of the animal and quantitative culture of the valvular vegetations. Such experiments have helped to eluci- date a hierarchy in the infectivity of the pathogens [13]. Adherence of circulating bacteria to the valvular endothelium/thrombotic vegetation is the most critical factor early in the pathogenesis of infective endocarditis [14,15].

Indeed, S. aureus, the VGS, and Enterococcus spp., which collectively account for the majority of cases of IE, do so specifically because of viru- lence factors that permit ligand-receptor interactions between bacterial surface components and constituents of damaged valves. However, the inoculum size (i.e., magnitude of the bacteremia) [13,16], as well as the duration of the bacteremia after inoculation, are also important determinants of infectivity [13].

Based on such models, antimicrobial prophylactic regimens should be predicted to be efficacious by interfering with one or more of these factors. A previously held belief was that

(3)

antibiotics prevented IE via elimination of the post-procedure transient bacteremia by killing the microorganisms before, as they entered, or while they were circulating in the bloodstream, before they seeded the endocardial surface. It seems unlikely, however, that any prophylactic agent could prevent the actual lodgement of circulating bacteria on a suitable nidus: seeding of the vegetation occurs within 30 minutes of the bacteria entering the circulation [17], while antibiotics usually require hours to exert their antibacterial effect [18]. The notion that prophylaxis is mediated by a bactericidal effect is the result of misinterpretation of negative blood culture results in earlier studies, which resulted from the continued elimination of the bacteria by the antibiotic after transfer of blood (and antimicrobial) to culture media. Indeed, animal [19,20] and human [21–24] studies with improved culture methods confirm that prophylaxis does not consistently and significantly reduce the incidence of post-procedure bacteremia. Therefore, the operative mechanism by which antibiotic prophylaxis is successful occurs by other means. Prevention of bacterial adherence has been proposed to explain the suc- cess of experimental prophylaxis. It was previously demonstrated that inhibitors of cell wall synthesis, such as β-lactams [25] and glycopeptides [20], have the capacity to decrease the adherence of bacteria to platelet-fibrin clots in vitro, possibly by inducing the release of lipoteichoic acid [26]. However, Moreillon and colleagues [27] elegantly demonstrated in the rat model of amoxicillin prophylaxis that inhibition of adherence was not an important mechanism, as the decrease was very marginal and did not prevent infection. Alternatively, successful prophylaxis is mediated by the ability of the administered antibiotic to facilitate elimination of bacteria subsequent to attachment to the vegetation. Studies have demonstrated that such an effect likely occurs by the prolonged inhibition of bacterial growth after inoculation. The determinants of the inhibitory effect include cha- racteristics of the organism (e.g., tolerance), the challenge dose (i.e., the ID₉₀, that is, the minimum inoculum producing IE in 90% of control animals), and the duration of time the serum concentration of the antibiotic remains above the MIC of the pathogen. Studies have shown that for inocula >ID₉₀, the longer the duration of growth inhibition, the greater the likelihood of successful prophylaxis [27–29].

Thus, when VGS or enterococci tolerant to

amoxicillin are inoculated into the rat model, single-dose prophylaxis with amoxicillin was efficacious only at the ID₉₀[16,30,31]. Against higher inocula, multiple doses of amoxicillin for VGS or amoxicillin and gentamicin for enterococci were necessary for successful prophylaxis [32]. Pharmacokinetic properties inherent in the administered antimicrobial assist in determin- ing the dosage scheme to maximize growth inhibition. For example, single-dose aminopenicillin prophylaxis for Enterococcus spp. is likely not effective because blood antibiotic levels are not sustained long enough completely to eliminate the bacteria from the vegetation, whereas single- dose teicoplanin was efficacious [33]. For organisms with demonstrated in vitro susceptibility, amoxicillin has a duration of inhibition of ≥ 10 hours [13]. These features identified from experimental models have thus allowed recommendations for prophylaxis in humans to be devised. What remains unclear, though, is the mechanism by which prolonged serum inhibitory activity eliminates bacteria adherent to vegetation. It had been postulated that growth-inhibited surface organisms would be susceptible to post-antibiotic leukocyte- enhanced opsonophagocytic activity. Animal studies [28], including a neutropenic endocarditis model [16], have demonstrated that poly- morphonuclear leukocytes do not play a role in eliminating bacteria adhered to the vegetation.

Therefore, the mechanism by which antibiotic prophylaxis is effective remains undefined.

Although the principle of prophylaxis dictates to administer the antimicrobial agent before commencement of the procedure, experimental studies have demonstrated that prophylaxis may also be effective if given shortly after the procedure. In the rat model, efficacy of prophylaxis could still be maintained if the antibiotic was administered within two hours of the bacteremia- inducing procedure [16]. Administration of antimicrobials at four to six hours post-procedure was not effective in preventing IE [16,34].

Also, although the dogma in the treatment of IE is to use a bactericidal antimicrobial regimen, this philosophy may not necessarily apply to IE prophylaxis, particularly given the lack of evidence that bactericidal properties mediate prophylaxis. In fact, animal studies have confirmed that while bactericidal antimicrobial agents are required for large inocula, bacteriostatic antimicrobial agents are effective for inoculum sizes

≤ ID90[35]—hence, the rationale for agents, such as the macrolides (e.g., clarithromycin [36]) and

(4)

lincosamides (e.g., clindamycin [37]) for penicillin allergic patients.

The applicability of the results from animal studies to humans remains debated. Major issues relate to the size of the inoculum used and the route of challenge. The bacteremia post- procedure in human is estimated to be <1 × 10² CFU/mL of blood [9], whereas in experimental models, the inocula used is in the order of 10⁶–10⁸ CFU/mL [38]. Such large inocula are required to ensure that IE consistently devel- oped in all (90%) of tested animals, but it may lead to an inaccurate model of disease.

Furthermore, most animal models are chal- lenged via the intravenous route to mimic a presumed mucosal micro-trauma-related bacteremia, again potentially introducing sources of error. Lastly, the experimental models used (i.e., rabbits, rats) may not reliably reproduce the pharmacokinetics of the antibiotics in humans, since these small animals clear drugs from their blood more quickly than humans [2].

Patients at Risk

The American Heart Association (AHA) [39], British Cardiac Society (BCS) [40], and French [41] guidelines stratify cardiac conditions into high- and moderate-risk categories, based on studies that have shown that certain types of structural heart disease are associated with higher risks of developing IE. Although the exact degree of risk for IE for certain cardiac lesions is difficult to assess, conditions deemed high-risk are inferred from the relative frequencies that particular cardiac lesions occur in a large series of patients with IE. For example, the incidence rates for IE are highest for patients with a previous history of native valve endocarditis (300–740/100,000 patient-years) and for patients with mechanical or bioprosthetic cardiac valves

(300–600/100,000 patient-years); these rates are approximately 60–185-fold higher than that of the general population [42]. Presumably, damaged valvular endothelium from a previous IE episode predisposes to subsequent nidus formation for a second episode. In the case of prosthetic valves, IE can occur by seeding of the foreign-body valvular apparatus. Patients with congenital cyanotic cardiac disease (i.e., single ventricle states, transposition of the great vessels, tetralogy of Fallot) also have higher incidence rates of IE, estimated at 100–200/ 100,000 patient-years; this represents a rate approximately 50-fold higher than that of the general population [42]. The increased incidence of disease in this group is likely related to turbu- lent, high-velocity flow and stagnant eddies from right-to-left shunts. It should be noted that strati- fication of cardiac conditions is also determined not only by risk of developing IE, but on the atten- dant morbidity or mortality should IE develop.

High- and moderate-risk categories are provided in Table 5.1.

Non-cyanotic congential heart disease includes conditions such as bicuspid aortic valve and coarctation of the aorta, as well as atrial septal defect (ASD), ventricular septal defect (VSD), and patent ductus arteriosus (PDA). Surgical repair of the latter three conditions has been reported to be associated with a negligible risk for IE (i.e., no greater risk than the general population). It should be noted, however, that the risk becomes negligible typically six months after surgical correction, provided that no other abnormality exists and no residual shunt is found by Doppler echocardiography, during which time endothelialisation of the material is complete [13,43].

Acquired valvular dysfunction includes aortic sclerosis, aortic stenosis (AS), aortic insuffi- ciency (AI), mitral stenosis (MS), and mitral regurgitation (MR). The prevalence of these valvulopathies increases with age. Of these, AS,

Table 5.1. Cardiac Conditions Associated with Increased Risk for IE High-risk:

1. Prosthetic cardiac valves (includes metallic, bioprosthetic, and homograft valves) 2. Previous endocarditis

3. Complex cyanotic congenital heart disease, e.g., single ventricle states, transposition of the great arteries, tetralogy of Fallot, double-outlet right ventricle 4. Surgically constructed systemic-to-pulmonic shunts/conduits

Moderate-risk

1. Most other congenital cardiac malformations, excluding atrial septal defect secundum and repaired atrial or ventricular septal defects 2. Acquired valvular dysfunction (e.g., rheumatic heart disease)

3. Hypertrophic obstructive cardiomyopathy

4. Mitral valve prolapse with mitral regurgitation or thickened leaflets on echocardiography

(5)

AI, MS, and MR result in abnormal high-velocity jet streams which can damage the endothelial lining and predispose to platelet aggregation and fibrin deposition on the valves, forming a nonbacterial thrombotic endocardial lesion.

These vegetations can act as a nidus for infection when seeded by circulating bacteremia.

Therefore, the ACC/AHA Guidelines for the management of patients with valvular heart disease [44] recommends that such patients, identified by physical examination or by echocardiography demonstrating at least moderate AS or MS or mild AI, receive IE prophylaxis.

Mitral valve prolapse (MVP), defined as a systolic displacement of all or part of a mitral valve leaflet at least 2 mm into the left atrium in a long-axis view on echocardiography, occurs in <5% of the general population [45]. MVP, however, is not uniformly associated with increased risk for IE. In fact, if auscultation reveals only the characteristic mid-systolic click and the valves are normal on echocardiography, the risk of IE in patients in this situation is negligible. However, if the valves are insuffi- cient, such that the characteristic murmur of MR is produced, or there is echoradiographi- cally demonstrable MR, prophylaxis is war- ranted [46]. If echocardiography demonstrates thickened, redundant mitral valve leaflets [45], such patients are also at increased risk for IE and prophylaxis should be administered [46].

In addition, male sex and age >45 years have been identified as predictors of increased risk for development of IE [45].

Procedures Producing Bacteremia

High-risk procedures, in this context, are those procedures associated with a high incidence of bacteremia, with “bacteremia” acting as a surrogate marker for IE risk. There is much contro- versy, however, about the role of invasive procedures, especially dental procedures, as the causative event leading to IE. The evidence for causality of odontogenic bacteremia is circum- stantial, based on a temporal relation between dental procedures and subsequent manifesta- tion of disease, and the identification of oral microflora (predominantly VGS, occasionally bacteria of the HACEK group) as the major pathogens. However, the mere presence of a temporal relation does not constitute proof of causation, particularly because of the influence

of reporting bias: dental procedures are extremely common (e.g., 62.8% of adults aged 18–64 reported ≥ 1 dental visit within the last year in 2002 [47]), whereas IE is relatively uncommon (e.g., 3.3 cases/100,000 population/

year in the United Kingdom, with similar figures for the United States [48] and France [41]).

Furthermore, identification of the same type of bacteria in the mouth and in cardiac vegetations supports the hypothesis that the offending pathogens derive from a mucosally lined source, but it again may be unfairly blaming dental procedures. There is no doubt that certain odontogenic procedures may occasionally cause transient bacteremias that lead to IE. However, it has been estimated that dental treatment causes no more than 4% of all cases of IE [49]. A population-based, case-control study by Strom and colleagues comparing 273 hospitalized adults with IE and 273 matched outpatient con- trols found that the calculated risk for IE was no higher in the first month after the dental treatment than after 2 or 3 months, demonstrating the absence of an association between the two events [50]. Pallasch, using a mathematical model, has estimated that the absolute risk rate for IE from a single dental treatment in the general population to be 1/14,258,714 dental visits [51]. Therefore, although it is convenient to think that gingival instrumentation with bleeding permits oral microflora to access the circulation and establish IE, the evidence that dental manipulation causes IE is weak. How then do the oral bacteria end up on the vegetation? The history of a “recent” dental procedure may, in fact, be a surrogate marker of poor oral hygiene.

Patients with poor oral hygiene are at increased risk for bacteremia in the absence of dental procedures, with the size of the inocula likely related to the degree of gingival inflammation [38,39]. Such transient bacteremia occurs with daily, trivial activities, such as chewing or tooth brushing. Guntheroth [49] devised a mathematical model to determine the cumulative exposure to bacteremia (CEB) resulting from

“physiologic” activities (e.g., mastication, brushing teeth), and compared it to that from a

“single dental extraction.” It was estimated that over a period of one hypothetical month, the physiologic CEB was 5,370 minutes, in contrast to 6 minutes for surgical CEB. The CEB method was modified by Roberts [38] to include the percentage prevalence of bacteremia related to the dentogingival manipulative procedure (p), the intensity of bacteremia (i, in colony-forming

(6)

units (CFU)/mL), the length of the bacteremia (t), and the frequency of bacteremia-inducing events estimated for a one-year period (f). The modified CEB (in CFU min/mL/year) for various activities were as follows: toothbrushing, 6,323;

flossing, 3,285; chewing, 3,285; single extraction of a permanent tooth, 0.014. To estimate the relative bacteremic challenge produced by one procedure versus another, the cumulative exposure index (CEI) was calculated, using the single deciduous molar extraction as the standard procedure, as it is widely recognized as causing a

“significant bacteremia” [38]. Roberts demonstrated that the CEI for toothbrushing twice a day is 154,219 times greater than that of an extraction.

He concludes that dental surgical procedures pose a low risk for IE. Rather, everyday procedures are much more likely (e.g., 8,000-fold higher risk) to cause transient episodes of low- grade bacteremias, that with time, results in a cumulative risk sufficient to cause IE. The mechanism by which this occurs is proposed to be via small movements of the tooth within the alveola, producing intermittent positive and negative pressures that cause microscopic gingival vascular damage, with subsequent aspiration of organisms into the circulation [38].

Further supporting the refutation of dental procedures as a major cause of IE are studies which raise doubt about the efficacy of pre-dental treatment antibiotic prophylaxis. In a nationwide, case-control study in the Netherlands, van der Meer and colleagues [52] estimated that the protective efficacy of chemoprophylaxis was 49% for first-ever IE occurring within 30 days of a procedure. The same group, in a prospective, population-based case study, demonstrated that medical and dental procedures cause only a small fraction of IE cases; furthermore, full compliance with prophylaxis might have prevented IE in 47 (17.1%) of 275 patients with late prosthetic or native valve IE involving a previously known cardiac lesion who underwent a procedure with an indication for prophylaxis.

For an incubation period of 30 days, prophylaxis might have prevented IE in 23 (8.4%) of these 275 patients, or 5.3% of all patients with endocarditis (i.e., total of 427 cases) [53]. The case-control study by Strom et al. [50] also challenges the use- fulness of IE prophylaxis, concluding that even if prophylaxis was 100% effective, it would reduce the incidence of IE by only 2.0 cases per 1 million person-years. A case-control study in France by Lacassin and colleagues [54] demonstrated that dental procedures were not associated with an

increased risk for IE, and that antibiotic prophylaxis provided a protective efficacy of only 46%, which was not statistically significant. These studies provide evidence suggesting that from a public health perspective, the routine use of antibiotic prophylaxis will only prevent a limited number of cases and is thus not justified.

However, three points need to be emphasized:

Firstly, some of the studies [50,54] still demonstrated an association between procedures in at- risk patients and the subsequent development of IE. Secondly, the studies were population-based, case- or case-control study design, raising the possibility of ecological fallacy in analysis inter- pretation, where the effect of antibiotic prophylaxis at the population level may be negligible, but may continue to be worthwhile for the indi- vidual patient [55]. Indeed, the study by van der Meer [52] admits that the small number of cases entered into the trial resulted in a small power that may have failed to detect a significant protective effect, and that there was the possibility that some subgroups may benefit from the use of prophylaxis. Lastly, case-control studies, with all their merits, are not the strongest level of evidence on which current medical decision making is based. These studies do, however, emphasize the importance of carefully identify- ing at-risk patients that will most benefit from prophylaxis. Furthermore, they underscore the need for more robust studies.

In the absence of a conclusive, prospective, randomized study, expert committees currently believe that prophylaxis should continue as recommended [39–41,56], despite the fact that it is an uncommon cause of IE, due to the high morbidity and mortality associated with this disease.

Although anaerobic bacteria are the principal components of the oral microflora and are released into the circulation after dental/oral procedures [21,57], they rarely cause IE. The predominant organisms of concern are the VGS, which are the targets for prophylaxis. A fundamental component of prophylaxis is good oral hygiene through daily, proper self-care and regular professional care. Antiseptic mouth rinses, either chlorhexidine- or povidone-iodine-based, may reduce the incidence and/or magnitude of bacteremia prior to dental procedures [58] and are recommended by the current AHA [39] and French [41] guidelines prior to invasive oral procedures to reduce the risk of IE. However, antimicrobial rinses do not permeate beyond 3 mm into the gingival sulcus and thus do not eradicate bacteria at the entrance into the

(7)

Table 5.2. Procedures Associated with Increased Risk for IE and Recommended Prophylaxis Regimens for Adults

Specific at-risk procedures for which Prophylaxis regimen for adults Category of procedures prophylaxis is recommended 1^stline 2^ndline Dental 1. Dental extractions 1. Amoxicillin 2 g po 1 h before Allergic to penicillin—

2. Periodontal procedures / gingival surgery procedure 1. Clindamycin 600 mg po 1hr before procedure Note: other dental procedures previously 2. If patient unable to take orally: OR

recommended include— Ampicillin 2 g IM/IV within 2. Cephalexin^*or cefadroxil^*2 g 30 min before procedure po 1 h before procedure

a. Dental implant placement and OR

replantation of avulsed teeth 3. Clarithromycin (or azithromycin) b. Root canal or surgery only beyond 500 mg po 1 h before

the apex procedure

c. Subgingival placement of antibiotic

fibres or strips Allergic to penicillin and unable

d. Initial placement of orthodontic bands to take orally—

(not brackets) 1. Clindamycin 600 mg IV within

e. Intraligamentous local anesthetic 30 min before procedure

injections OR

f. Prophylactic cleaning of teeth or implants 2. Cefazolin^*1 g IV within 30 min

where bleeding is anticipated before procedure

Respiratory 1. Tonsillectomy or adenoidectomy 2. Bronchoscopy with a rigid bronchoscope 3. Surgery involving the respiratory mucosa Esophagus 1. Sclerotherapy of esophageal varices

2. Esophageal stricture dilatation

Gastrointestinal 1. Endoscopic retrograde cholangiography High-risk—Ampicillin 2 g IM/IV + High-risk and allergic to β- with biliary obstruction gentamicin 1.5 mg/kg (max: lactams—Vancomycin 1 g IV over 2. Biliary tract surgery 120 mg) within 30 min before 1-2 h + gentamicin 1.5 mg/

3. Surgery involving the intestinal mucosa procedure THEN kg (max: 120 mg); complete Ampicillin 1 g IM/IV (OR infusion within 30 min before amoxicillin 1 g po) 6 h later procedure

Genito-urinary 1. Cystoscopy Moderate-risk— Moderate-risk & allergic to β- 2. Urethral dilatation 1. Amoxicillin 2 g po 1 h before lactams—

3. Prostatic surgery procedure Vancomycin 1 g IV over 1–2 h;

2. If patient unable to take orally— complete infusion within 30 min Ampicillin 2 g IM/IV within before

30 min before procedure

*Cephalosporins should not be used in patients with type 1/immediate-type hypersensitivity to b-lactams.

Routes of administration: po = orally; IM = intramuscularly; IV = intravenously.

systemic circulation [59], raising the need by some for more supportive evidence of benefit.

Systemic antibiotic prophylaxis is recommended for at-risk patients (see Table 5.2).

Prophylaxis is recommended for procedures associated with significant bleeding [3,13,39]. As well, it is recognized that unanticipated bleeding may occur on occasion in patients who did not receive prophylaxis prior to the procedure; in these cases, experimental data suggests that the appropriate pre-procedure regimen can still be administered within two hours of the procedure with similar efficacy [13,39]. Interestingly, however, visible bleeding may not be a clinically rel- evant tool, as a previous study has demonstrated that bleeding is a poor predictor of odontogenic bacteremia [38]. In cases where multiple consec-

utive dental interventions are required, repeated prophylaxis is also required. Because repeated single-dose antibiotic administration may select for resistant organisms which persist in the mouth, multiple procedures are recommended to be carried out in one sitting (if possible) or separated by 9–14 days [39,41].

Streptococcal bacteremia can also occur via manipulation of other mucosal surfaces lining the upper respiratory tract (e.g., tonsillectomy [60–62], mastoidectomy [63], septoplasty [64]).

Although the use of a rigid bronchoscope is sug- gested to be a potential bacteremic-inducing procedure via mucosal damage and for which prophylaxis is recommended [39,40,56,65], there is no literature to support this opinion. In fact, one prospective nonrandomized clinical

(8)

study in 25 children undergoing diagnostic rigid tracheobronchoscopy for airway assessment demonstrated no cases of bacterial growth in blood cultures [66]. Fiberoptic bronchoscopy, previously thought to be benign with five studies (291 patients) demonstrating a procedure- induced bacteremia rate <1% [67] and for which prophylaxis is not recommended [39,56,65], may actually be associated with higher rates.

Yigla and colleagues [67] demonstrated a bacteremia rate of 6.5% in a prospective study of 200 consecutive patients that underwent fibre- optic bronchoscopy without either pulmonary infection or an unusually high rate of invasive procedures. If additional studies can support this finding, it may have implications in future revisions of IE prophylaxis guidelines.

The esophageal procedures with the highest associated bacteremia rates are sclerotherapy of esophageal varices and esophageal dilation of a stricture [68,69]. Earlier studies have demonstrated rates of 31% for sclerotherapy (61 patients) and 45% for dilation (59 patients), in which the majority of organisms were VGS [68].

More recent prospective studies support these rates. Zuccaro and colleagues [70] performed blood cultures before and after stricture dilation in 103 patients without valvular heart disease and in a control group of 50 patients undergoing upper endoscopy without dilation. They demonstrated that 21% (22/103) of patients undergoing dilation had positive blood cultures, with VGS as the predominant isolate. Among 100 procedures in 86 patients undergoing esophageal dilation by Nelson et al. [71], 22 (22%) were associated with a positive post-dilation blood culture. Although these episodes of bacteremia post-endoscopy are short lived (i.e., typically <30 minutes), their clinical significance is unclear (as it is with other post-procedure bacteremias). One prospective comparative study randomizing 39 patients to prophylaxis (i.e., cefotaxime, 19 patients) or no antibiotic (20 patients) revealed a significant reduction in post-procedure bacteremic episodes in the group receiving antibiotic (5.3%

vs. 31.6%, respectively; P = .04)) [72]. However, a recent review of the infectious disease complications of GI endoscopy has revealed only two cases of IE after sclerotherapy have been reported, one involving a prosthetic valve (despite prophylactic administration of appropriate antibiotics) and another on a native valve [69]. Nonetheless, current guidelines continue to recommend prophylaxis for these procedures

[39,41,73]. Endoscopic variceal ligation (EVL,

“banding”) has replaced sclerotherapy as the procedure of choice in the management of varices because of its greater efficacy and fewer associated complications. In a historical cohort study comparing the rates of transient bacteremia between the two procedures, positive blood cultures occurred more frequently in the sclerotherapy group (17.2%) than in the ligation group (3.3%, P < 0.03) [74]. A review of seven studies addressing this issue, including the one mentioned, reports bacteremia rates associated with EVL ranging from 0% to 25%, with a mean frequency of 8.8% [69]. The attributable risk of IE to endoscopic variceal ligation is unknown, as no cases have currently been reported in the English literature.

Endoscopic retrograde cholangiopancreato- graphy (ERCP) has become a commonly performed procedure. The diagnostic and therapeutic utility of ERCP has been well demonstrated for a variety of disorders, including the management of biliary obstruction, predominantly due to choledocholithiasis or biliary malignancies. The rate of bacteremia after contrast injection or instrumentation of unob- structed pancreatic or bile ducts ranges from 0%

to 15% (mean frequency of 6.4%) [69]. Biliary obstruction, however, may lead to infection of the biliary system with a variety of organisms.

Although the predominant organisms are Gram-negative bacillary enterics (e.g., E. coli, Klebsiella spp.) [75,76], which are common causes of cholangitis/biliary sepsis, they are uncommon causes of IE, although they may cause disease in high-risk patients (e.g., those with prosthetic valves). The major organisms from an infected biliary tree that can cause bac- teremia with the potential for IE are Enter- ococcus spp. and VGS [75]. The enterococci are particularly more common among patients with previous biliary endoprosthesis [76]. Instru- mentation of an obstructed biliary system has resulted in bacteremia rates as high as 26.5%

(mean 18.0%) [69], hence the rationale for prophylaxis. Although earlier studies provided some evidence that prophylaxis may reduce the incidence of post-ERCP bacteremia [77,78], a meta-analysis by Harris and colleagues [79] that reviewed five prospective, randomized placebo- controlled trials failed to show such a benefit among patients who received prophylaxis, argu- ing against the routine prophylactic use of antibiotics prior to ERCP to reduce bacteremia.

(9)

This is not to say, however, that antibiotics should not be used in patients with known cholangitis. As well, because the meta-analysis excluded two studies where patients received antibiotics before and after the ERCP, it is possible that continuation of the prophylaxis after the procedure may reduce bacteremia. Therefore, such a regimen continues to be recommended for patients with biliary obstruction and high- risk for IE [39–41,73]. A similar rationale exists for surgery on the biliary and gastrointestinal tracts [39–41].

Endoscopic ultrasound (EUS) is a relatively new procedure. One of its greatest benefits is the ability to perform fine-needle aspiration (FNA), the two procedures referred to as EUS-FNA.

EUS-FNA has been used to aspirate fluid from cystic lesions, pseudocysts, and fluid collections for both diagnostic and therapeutic purposes [80]. The frequency of bacteremia as a compli- cation of EUS and EUS-FNA has been prospec- tively studied in 3 separate trials, which included approximately 250 patients [81–83]. These studies did not find a statistically significant increase in the rate of bacteremia when compared with that seen at upper endoscopy. Based on these data, prophylactic antibiotics are not recommended for FNA of solid masses and lymph nodes [80]. Some experts recommend prophylactic antibiotics as well as 48 hours of antibiotics after the procedure for EUS-FNA of the perirectal space [80]. EUS-FNA of cystic lesions appears to carry an increased risk of febrile episodes and possibly sepsis and, therefore, war- rants prophylactic antibiotics, as well as a short postprocedure course [80].

Colonoscopy has a surprisingly low rate of bacteremia (2–5%) [10,69,84], most commonly with organisms that are not typically causes of IE. Therefore, antibiotic prophylaxis is not recommended for this procedure, including when it involves biopsy or polypectomy [85].

Genitourinary (GU) instrumentation is necessary for the diagnosis and treatment of benign and malignant urological diseases. However, instrumentation and catheterization of the GU tract is also the leading cause of nosocomial urinary tract infections (UTIs) [86]. Less frequently, bacteremia can result from these interventions, the rates varying with different procedures. Development of bacteremia directly attributable to the GU procedure typically occurs after colonization of the urine. As such, the majority of studies on the use of prophylac-

tic antibiotic regimens prior to GU interventions have assessed the efficacy in preventing UTIs.

There have been only a few studies that have assessed the efficacy in preventing bacteremia, reflecting the infrequent occurrence of this com- plication. When bacteremia occurs, the clinical manifestations range from asymptomatic, to transient fever, to septicemia/urosepsis. IE due to manipulation of the GU tract is extremely uncommon, but has been reported [87–89]. As such, the evidence for IE prophylaxis in GU procedures is scant and is based largely on the efficacy in preventing bacteremia, as well as on expert opinion.

As with lower gastrointestinal procedures, GU procedures will mostly produce bacteremia with Gram-negative organisms (e.g., E. coli, Klebsi- ella spp. [90–92]), which are common causes of urosepsis but are uncommon causes of NVE.

These organisms may, however, cause IE in high-risk patients (e.g., those with prosthetic valves). Of the organisms arising from the native GU tract, the predominant ones that may cause NVE are Enterococcus spp. and the VGS [92].

Although the risk that any particular patient will develop endocarditis is low, the rate of bacteremia following invasive urinary tract instrumentation is high in the presence of bacteriuria.

For example, cystoscopy, urethral dilation, and transurethral resection of the prostate (TURP) in the presence of bacteriuria precipitated bacteremia at rates of 25% [93], 40% [92], and 52%

[92], respectively. This perioperative bacteremia is usually transient and symptomless—in as many as ~6% of cases in one study [94]—

though it may progress to perioperative septicemia. With this in mind, sterilization of the urinary tract with antimicrobial therapy in patients with bacteriuria should be attempted prior to elective procedures [93,95]. Such intervention has been shown to reduce the risk of septicemia [96]. Whether it also reduces the risk of IE is unknown. However, in a study of 15 non-catheterized patients with sterile urine, cystoscopy resulted in post-procedure bacteremia in 13% of patients [97], which can theo- retically result in IE in at-risk patients. As well, the incidence of post-procedure bacteremia after transurethral procedures (i.e., TURP, transurethral resection of bladder tumour/

TURBT) ranged from 30% to 45% in three prospective, comparative studies [98–100], which was reduced by approximately 80–90%

with appropriate antimicrobial prophylaxis

(10)

[93]. These studies were marked, however, by relatively high rates of bacteriuria in both com- parison groups [93], which accounts for the high rates of bacteremia in the absence of prophylaxis. In a meta-analysis of ten randomized controlled trials of antibiotic prophylaxis for TURP in men with sterile urine (i.e., preoperative urine specimen containing < 1 × 10⁵CFU/mL), a significant decrease in the frequency of postopera- tive bacteremia was noted with the intervention, albeit with lower baseline rates (4% vs. 1%, risk difference of −0.02, 95% confidence interval of

−0.04–0.00) [101]. The rate of bacteremia after combined cystoscopy and transrectal biopsy of the prostate was 73% in one study [97]. Hence, mono-antimicrobial prophylaxis (e.g., aminopenicillins or glycopeptides) is recommended for moderate-risk patients prior to these urological procedures to target the above mentioned Gram- positive organisms. For high-risk patients, combination therapy targeting Gram-positive and Gram-negative flora is recommended.

Antimicrobial Prophylaxis

Because VGS are felt to be the predominant pathogens potentially to cause IE after dental/oral, respiratory, and esophageal procedures, aminopenicillins are the recommended prophylaxis. In the past, VGS were nearly uniformly susceptible to penicillin and other β-lactams, as well as to lincosamides and macrolides [102]. Therefore, the current AHA guidelines on IE prophylaxis, which were published in 1997 [39], recommend the use of amoxicillin (ampicillin if the patient is unable to tolerate oral intake). Amoxicillin was recommended over penicillin because it is better absorbed from the GI tract and because it provides higher and more sustained levels [39]. In humans, the elimination half-life of amoxicillin is 50–60 minutes [103]. Clindamycin or macrolides are alternatives in those unable to tolerate β-lactams. A contemporary review of the antimicrobial susceptibility of VGS demonstrated that amoxicillin at a concentration of ≤ 0.5 µg/mL inhibited 87%, 64%, and 100% of isolates in the S. sanguis, S. mitis, and S. milleri groups, respec- tively, as well as two of the three isolates in the S. salivarius group [104]. Hence, the use of amoxicillin as a prophylactic regimen was justified. However, several studies have since demonstrated increasing rates of VGS isolates from oropharyngeal specimens [105] and blood-

stream infections [102,106–109] that are not susceptible to penicillin, macrolides, or lincosamides. Furthermore, resistance to these antibiotics can occur with repeated prophylaxis doses for serial procedures distributed closely in time [39,41]. Therefore, continued monitoring of such resistance patterns is mandatory, and modifications of future guidelines may be necessary. Until such time, amoxicillin remains the recommended prophylaxis regimen for the above-mentioned procedures. When comparing the AHA guidelines from those of Europe (BSC, French), differences in amoxicillin dose is seen.

The latter recommend a single 3-g oral dose, which produces serum levels above the MIC of most oral streptococci for a period of 6–14 hours [110]. The AHA proposes 2-g, instead of 3-g, because the serum kinetics produced by the two different doses are very similar, although the lower dose is associated with fewer side effects [111]. For patients with a history of penicillin allergy, clindamycin remains appropriate. Alter- natives include macrolides, such as clarithromycin or azithromycin, which have demonstrated efficacy in experimental models and have convenient dosing regimens, although they are more expensive. Cephalosporins also have demonstrated efficacy, but should not be used in patients with a history of type 1 (immediate-type/anaphylaxis) hypersensitivity reaction to β-lactams. For patients unable to take medica- tion orally, intravenous regimens are recommended, and administration of the full dose should be completed within 30 minutes of the procedure.

For procedures involving the biliary system or the gastrointestinal or genitourinary tracts, the predominant pathogen of concern is Enterococcus spp. Previous studies have repor- ted that among cases of enterococcal IE, ~40%

were associated with a recent gastrointestinal or genitourinary procedure (i.e., within 2–6 weeks) [28]. Enterococci however, are notoriously more resistant than VGS, with typically higher MICs to β-lactams [112]. Thus, after administration of amoxicillin, the corresponding serum levels fall below the MIC of enterococci sooner than for VGS, resulting in a decreased period of bacterial growth inhibition. To overcome this issue in high-risk patients, a second dose of the β-lactam is currently recommended six hours after the first dose to ensure prolongation of adequate serum levels and to enhance protective efficacy.

The rationale for the combination of amoxicillin and gentamicin is based on the rat model of

(11)

Enterococcus IE, in which administration of both agents was necessary for successful prophylaxis against inocula >ID₉₀[32]. Alternatively, administration of a single dose of vancomycin (in con- junction with gentamicin) can be used in high-risk patients unable to tolerate β-lactams.

The evidence for this recommendation derives from experimental studies in which vancomycin demonstrated prolonged serum half-life, producing serum levels greater than MIC for a longer period of time (compared to ampicillin- based regimens), which resulted in significantly greater area under the curve (AUC) and serum inhibitory activity, and more consistent protective effect [28]. Because of vancomycin’s pharmacokinetics, a second dose is not considered necessary. For moderate-risk patients, the second dose of aminopenicillins is optional.

Reasons Against Prophylaxis

Since IE is potentially fatal, prophylaxis seems reasonable. The benefit of giving antibiotic prophylaxis to otherwise healthy people, however, should outweigh its risks. The major complications associated with administration of prophylaxis include allergic reactions, toxic side effects of antimicrobials, adverse interactions with other drugs, and development of resistant organisms.

The most significant adverse event associated with the penicillins is hypersensitivity reactions, which can range from a troublesome rash to life- threatening anaphylaxis. Previous studies that have compared the rates of IE-associated deaths to the rates of deaths from antibiotic-induced anaphylaxis have questioned the benefit of prophylaxis. In a quantitative analysis of published data on prophylaxis in patients with mitral valve prolapse (MVP), Bor and Himmelstein [113] cal- culate that among 10 million patients with MVP undergoing a dental procedure, an estimated 47 nonfatal cases and 2 fatal cases of IE would occur if no prophylaxis were given, compared to 5 cases of IE and 175 deaths due to drug reactions if all patients were given prophylaxis with a penicillin.

Similarly, Tzukert and colleagues [114] demonstrated that patients receiving penicillin/amoxicillin propylaxis to prevent IE are five times more likely to die from anaphylaxis to the drug than from IE, with estimated rates of 1.36 deaths versus 0.26 deaths per million population,

respectively. These studies were conducted in the mid-1980s, and national guidelines have since been revised to tailor prophylaxis to at-risk patients. No study has since demonstrated whether the risk–benefit ratio has been modified by the latest recommendations. Nonetheless, the potential for adverse drug reactions must always be borne in mind. Such a consideration should also include non-allergic toxicities (e.g., amino- glycoside-induced nephrotoxicity), as well as potential drug–drug interactions.

An emerging problem resulting from inappro- priate use of antimicrobial agents is the develop- ment of C. difficile-associated disease (CDAD).

C. difficile is the most common cause of infec- tious diarrhea among hospitalized patients. It is well-documented that recent antibiotic use (e.g., within 42 days [115]) predisposes to acquisition of C. difficile. Essentially all antibiotics have been associated with risk for CDAD, including those recommended for IE prophylaxis. In a meta- analysis by Bignardi [116], use of ampicillin or amoxicillin was associated with a pooled odds ratio of 3.7 for acquiring disease (95% CI:

2.6–5.5), while the rates for clindamycin, 1^st-generation cephalosporins, and vancomycin were 9 (6.3–12.9), 2.6 (1.8–3.7), 3.1 (1.8–5.2), respectively. Development of CDAD leads to prolonged hospitalizations [117,118]. It can also be associated with severe disease (i.e., megacolon, perfo- ration, colectomy, shock requiring vasopressor therapy, or death within 30 days after diagnosis) [119]. In certain geographic areas, CDAD is associated with increased mortality rates, with a one- year cumulative attributable mortality of 17%

[117]. Development of CDAD following antibiotic prophylaxis for dental procedures has been reported [120], as it has after single doses of antibiotics for other procedures [121,122].

Emergence of CDAD emphasizes the need to weigh the risks versus the benefits of antibiotic prophylaxis.

An additional concern from the large-scale use of IE prophylaxis is the development of antimicrobial resistance. In healthy human vol- unteers, administration of repeated doses of amoxicillin was followed by emergence of resist- ant VGS from the oral flora [123]. A case of S.

mitis IE developing despite seemingly-appropri- ate prophylaxis has been reported in a patient who received two recent courses of amoxicillin for dental procedures [124]. In the neutropenic cancer patients, exposure to previous β-lactams was associated with an increased risk of bloodstream infection (non-endocarditis) with

(12)

β-lactam-resistant VGS [125,126]. Previous exposure to antibiotics has also permitted the emergence of resistant enterococci [127] and S. aureus [128,129]. Consequently, judicious use of antibiotics, in general, is advocated, and administration of antimicrobial prophylaxis should not be done indiscriminately, but tai- lored to those specifically at-risk for disease.

Emerging Issues

The current recommendations for IE prophylaxis are based on an epidemiology in which VGS were the predominant pathogens. Recent studies have demonstrated that S. aureus has become the major cause of IE [130]. An increasing propor- tion of cases of S. aureus bloodstream infection and IE is acquired nosocomially or nosohusially (i.e., health-care-associated) [130–132], due to increasing use of intravascular devices (e.g., central venous catheters, dialysis catheters, prosthetic vascular grafts, pacemakers/ defibril- lators). These devices can also permit coagulase- negative staphylococci (CoNS, e.g., S. epidermidis) to establish endovascular infections. Indeed, the incidence of CoNS IE is also increasing [133]. The existing aminopenicillin-based prophylaxis recommendations are not likely to be effective in preventing S. aureus IE, based on in vitro susceptibility testing in which <5% of clinical isolates are inhibited by penicillin [134–136].

Similarly, they are not expected to be effective against CoNS. There are currently no national guidelines regarding IE prophylaxis for the above-mentioned procedures. The recommendations that exist recommend prophylaxis to minimize the risks of intraoperative contami- nation and surgical site infection [137]. Typically a first-generation cephalosporin directed prima- rily against staphylococci is administered in the peri-implantation time period for clean- contaminated procedures, and only for a short duration (e.g., a few doses) [137]. This approach, however, may not be adequate to prevent bacteremia. For devices in which a portion remains external to the patient, and thus provides a persistent portal of entry, the brief administration of the peri-procedure prophylaxis is cer- tainly not sufficient to prevent bacteremic episodes that may occur during the lifespan of the implanted device. In particular, the use of central venous catheters (CVCs) has emerged

as a major risk factor for bacteremia and IE [132]. Consequently, health-care-associated IE (HA-IE), defined as acute IE occurring 48–72 hours or more post-admission to hospital and/or IE directly relating to a hospital-based procedure performed during a previous hospital stay within eight weeks of admission, currently accounts for approximately 7.5–29%

of all cases of IE seen in tertiary hospitals [138]. As such, modification of IE prophylaxis recommendations is required to address this changing epidemiology. One intervention which may be particularly useful for preventing CVC colonization, and therefore may minimize the risk of bacteremia and IE, is the antibiotic lock technique. This technique consists of filling and closing of the catheter lumen with a high-concentration antibiotic solution that acts locally to eradicate catheter- associated bacteremia, but that allows the side effects and toxicity associated with systemic administration of antibiotic to be avoided.

Future studies are required before such intervention can be recommended.

Conclusion

Guidelines exist to assist clinicians in stratifying their patients’ risk of IE with regard to various procedures. Unfortunately, most of the recommendations are not based on robust, scientific evidence, but, instead, are consensus expert opinion. In addition, emergence of antimicrobial resistance and a changing epidemiology of IE will likely necessitate revision of current guidelines.

Key Points

1. Guidelines exist for antibiotic prophylaxis against infective endocarditis (IE). There is lit- tle robust clinical evidence supporting proof that antibiotic prophylaxis decreases the immediate subsequent risk for IE. The strength of the evidence rests on animal studies, which may or may not accurately reflect human disease, as well as on expert opinion. Nonetheless, a priori algorithms have been proposed for the health-care practitioner, based on patient risk factors for disease as well as the likelihood of bacteremia from a given procedure.

2. The mechanism(s) by which antibiotics affect prophylaxis remain unclear, but may involve